Network simulation is used to design various kinds of networks, simulate and then analyze the effect of various parameters on network performance. Network simulation tools can test scenarios that might be particularly difficult or expensive to emulate using real hardware.
In a simulation of a radio network, where a message is transmitted wirelessly from one node to another, a perfect communication may be modeled. In a perfect communication, every message is received without error or delay. Although a perfect communications model significantly simplifies the simulation, it does not reflect the actual conditions that wireless communications encounter in the real world. Thus, it can introduce significant inaccuracies in simulation results.
For a more accurate simulation, communication effects may be incorporated. Communication effects refer to communications within a simulation environment in a manner that is consistent with effects experienced in the real world. For instance, messages might be dropped, corrupted, delayed, jammed, or blocked by terrain, or they may be affected by weather, network congestion, etc.
To incorporate realistic communication effects, calculations are performed on each simulated radio transmission to evaluate the effects of certain characteristics (e.g., location of receiver, blockage of line of sight, effects of network protocols) on each transmitter and receiver. Each message is either delivered with realistically calculated errors, noise, and delays to each separate receiver or it is not delivered at all (that is, dropped).
The communication effects are simulated by performing calculations inline at the time they occur in the simulation. For each separate radio transmission that occurs in a simulation, a network simulator tool is invoked to perform all communication effects calculations at that time for that message, and then the tool determines whether the message was dropped or delivered.
There are problems with simulating communication effects in large networks. In simulations of low latency networks, delivery decisions can take too long to compute. The time taken to simulate a transmission might take considerably longer than the time required to deliver the decision. If target/contact calculations are based on a (realistic) low latency (as is the case with cockpit simulators), the entire simulation can be rendered invalid or unworkable.
Drop decisions can cause additional problems. If a network simulator cannot distinguish between a message that is slow and a message that will never arrive, it will wait until the message arrives or a timeout occurs (whichever occurs first). By waiting, the receivers tie up buffers.
Network scalability is also a problem. Each communication effect can require a large amount of computations. Moreover, the number of computations per transmission increases as higher fidelity communication effects is needed. Thus, high fidelity simulations might be achievable for a hundred nodes, but not for thousands of nodes. Or, if the number of computations per transmission is held the same as the number of simulated nodes increases, the communication effects fidelity is reduced.